diggis_tennis
Newbie level 3
Re: HELP Needed for Matlab simulation for rate 1/3 encoder d
Hi,
I am working on this project where in I have to simulate (matlab, C) a rate 1/3 encoder decoder ( soft or hard) to get minimized gap to capacity over a AWGN channel compared to un coded BPSK. Also I have been given a sample skeleton in which I just need to modify encoder and decoder part else I can write a new program is I want. Can some one please help me here. I know its a simple code but I am a newbie and I have a coding phobia. Thanks. The code can be anything ldpc, convolutional but the performance should be good in terms of the capacity thing. Also he gave a hint to use concatenated Reed solomon with convolutional codes.
Added after 4 minutes:
Here is the skeleton code
L = 1e4; % message length
R = 1/3; % code rate
dbs = -1:10; % SNR per bit, in dB
trials = 1e3; % number of trials to perform
%------------
errs = 0*dbs;
EbN0 = 10.^(dbs/10);
sigs = 1./sqrt(2*R*EbN0);
ber0 = logspace(-6,-2.1,81);
ber1 = logspace(-6,-0.99,81);
db0 = 10*log10((2.^(2*R*(1+log2((ber0.^ber0).*(1-ber0).^(1-ber0))))-1)/(2*R));
db1 = 20*log10(erfinv(1-2*ber1));
for trial = 1:trials,
m = round(rand(1,L)); % message vector
%--------- ENCODER: (3, 1) repetition code --------%
c = reshape([m; m; m],1,3*L);
%--------------------------------------------------%
%verify code rate!
if trial==1,disp(['Measured R = ',num2str(length(m)/length(c))]);end;
noise = randn(1,length(c));
for i=1:length(dbs),
r = 2*c - 1 + sigs(i)*noise;
%--- DECODER: soft (3,1) repetition decoder ---%
mhat = sum(reshape(r,3,L))>0;
%----------------------------------------------%
errs(i) = errs(i) + sum(mhat ~= m);
end;
ber = errs/(L*trial);
semilogy(dbs, ber,'o-', db0, ber0,':', db1, ber1,':');
xlabel('SNR per bit, E_b / N_0 (dB)');
ylabel('Bit-Error Rate');
axis([-1 10 1e-6 1])
title(['After ',num2str(trial),' trials (',num2str(L*trial),' msg bits)']);
drawnow;
end;
Hi,
I am working on this project where in I have to simulate (matlab, C) a rate 1/3 encoder decoder ( soft or hard) to get minimized gap to capacity over a AWGN channel compared to un coded BPSK. Also I have been given a sample skeleton in which I just need to modify encoder and decoder part else I can write a new program is I want. Can some one please help me here. I know its a simple code but I am a newbie and I have a coding phobia. Thanks. The code can be anything ldpc, convolutional but the performance should be good in terms of the capacity thing. Also he gave a hint to use concatenated Reed solomon with convolutional codes.
Added after 4 minutes:
Here is the skeleton code
L = 1e4; % message length
R = 1/3; % code rate
dbs = -1:10; % SNR per bit, in dB
trials = 1e3; % number of trials to perform
%------------
errs = 0*dbs;
EbN0 = 10.^(dbs/10);
sigs = 1./sqrt(2*R*EbN0);
ber0 = logspace(-6,-2.1,81);
ber1 = logspace(-6,-0.99,81);
db0 = 10*log10((2.^(2*R*(1+log2((ber0.^ber0).*(1-ber0).^(1-ber0))))-1)/(2*R));
db1 = 20*log10(erfinv(1-2*ber1));
for trial = 1:trials,
m = round(rand(1,L)); % message vector
%--------- ENCODER: (3, 1) repetition code --------%
c = reshape([m; m; m],1,3*L);
%--------------------------------------------------%
%verify code rate!
if trial==1,disp(['Measured R = ',num2str(length(m)/length(c))]);end;
noise = randn(1,length(c));
for i=1:length(dbs),
r = 2*c - 1 + sigs(i)*noise;
%--- DECODER: soft (3,1) repetition decoder ---%
mhat = sum(reshape(r,3,L))>0;
%----------------------------------------------%
errs(i) = errs(i) + sum(mhat ~= m);
end;
ber = errs/(L*trial);
semilogy(dbs, ber,'o-', db0, ber0,':', db1, ber1,':');
xlabel('SNR per bit, E_b / N_0 (dB)');
ylabel('Bit-Error Rate');
axis([-1 10 1e-6 1])
title(['After ',num2str(trial),' trials (',num2str(L*trial),' msg bits)']);
drawnow;
end;
